使用gprof进行多线程程序性能分析

最新推荐文章于 2024-08-25 15:31:37 发布

原创最新推荐文章于 2024-08-25 15:31:37 发布 · 1k 阅读

0 ·

CC 4.0 BY-SA版权

本文为博主原创文章，未经博主允许不得转载。本博客持续更新，欢迎关注。

工具软件同时被 2 个专栏收录

41 篇文章

订阅专栏

性能优化

12 篇文章

订阅专栏

本文介绍如何使用GNU Profiler (gprof) 工具有效分析多线程应用程序的CPU使用情况。通常，gprof在某些内核下如Linux，只能分析主线程，但通过设置ITIMER_PROF定时器并将其传递给所有派生线程，可以解决这一问题。文章提供了一个优雅的解决方案，即为pthread_create实现一个包装器，并附带了预加载库中使用的代码示例。

HOWTO: using gprof with multithreaded applications

What is gprof?

gprof is the GNU Profiler, a tool used when tracking which functions are eating CPU in your program. Anyway, you should already be familiar with it if you got interested in this page.

One problem with gprof under certain kernels (such as Linux) is that it doesn’t behave correctly with multithreaded applications. It actually only profiles the main thread, which is quite useless.

Workaround

There is an easy, but surprisingly not very widespread fix for this annoying gprof behaviour. Basically, gprof uses the internal ITIMER_PROF timer which makes the kernel deliver a signal to the application whenever it expires. So we just need to pass this timer data to all spawned threads.

Example

It wouldn’t be too hard to put a call to setitimer in each function spawned by a thread, but I thought it would be more elegant to implement a wrapper for pthread_create.

Daniel Jönsson enhanced my code so that it could be used in a preload library without having to modify the program. It can also be very useful for libraries that spawn threads without warning, such as libSDL. The result code is shown below and can be downloaded (gprof-helper.c):